Lambert Guidance in Astrodynamics

• Lambert Guidance is a method to compute spacecraft velocity vectors for transferring between two positions within a prescribed time under a central gravitational field.
• It employs universal-variable time-of-flight formulations with Stumpff functions, using algorithms like Newton–Raphson and Householder to achieve rapid, robust convergence.
• This technique underpins key applications in orbit determination, rendezvous, and real-time trajectory optimization in modern space missions.

Lambert guidance addresses the impulse control of a spacecraft or vehicle maneuvering between two position vectors under a central gravitational field, with the boundary condition of a prescribed transfer time. It is foundational to trajectory design, orbit determination, rendezvous, and interception in astrodynamics, where the solution—Lambert’s problem—yielding the required initial and final velocities for transfer is tractable in closed-form for Keplerian orbits. Several solution algorithms have matured, grounded in the universal-variable time-of-flight formulation and direct geometric analysis, each optimized for speed, accuracy, and robustness across boundary scenarios including degenerate and multi-revolution paths (Sangrà et al., 2021, &&&1&&&, Easton et al., 2021).

1. Formal Statement of Lambert’s Problem

Given a central body of gravitational parameter $\mu=GM$, and two position vectors $\mathbf{r}_1$ at $t_1$ and $\mathbf{r}_2$ at $t_2 = t_1 + \Delta t$, the boundary-value problem seeks velocity vectors $\mathbf{v}_1$ and $\mathbf{v}_2$ such that the Keplerian conic passing through both positions is traversed in precisely $\Delta t$. The orbital transfer angle $\theta$ is defined as the angle between $\mathbf{r}_1$ and $\mathbf{r}_2$; both “short-way” ($0<\theta<\pi$) and “long-way” ($\pi<\theta<2\pi$) solutions are admissible. Multi-revolution solutions augment $\theta$ by integer multiples of $2\pi$ (Gatland, 2021, Easton et al., 2021).

2. Universal Variable Time-of-Flight Formulation

The modern solution approach parameterizes the family of Keplerian conics by a universal anomaly $\chi$, facilitating elliptic, parabolic, and hyperbolic regimes via Stumpff functions:

• $C(\chi) = 1 - \chi^2/2! + \chi^4/4! - \ldots$,
• $S(\chi) = 1/3! - \chi^2/5! + \chi^4/7! - \ldots$.

The geometry constant $A = \sin \theta \sqrt{\,r_1 r_2\,/(1-\cos \theta)}$ and the auxiliary quantity $y = r_1 + r_2 + (A/\chi)\bigl(1 - C(\chi)\bigr)$ define the closed-form time-of-flight relation:

$$\Delta t(\chi) = \frac{1}{\sqrt{\mu}} \left[ \chi^3 S(\chi) + A\sqrt{y} \right].$$

The transcendental equation $$F(\chi) \equiv \Delta t(\chi) - \Delta t_{prescribed} = 0$$ is solved for $\chi^*$, selecting the branch consistent with revolution count and transfer direction (Sangrà et al., 2021).

3. Primary Solution Algorithms and Their Properties

Newton–Raphson Universal-Variable Method (Bate)

• Initializes at $\chi_0=0$ (parabolic seed) or via bracketing; iterates with

$$\chi_{k+1} = \chi_k - \frac{F(\chi_k)}{F'(\chi_k)},$$

where the derivative $F'(\chi)$ is explicit in terms of $C(\chi)$, $S(\chi)$, $y$, and $A$.

• Converges in $5$–$10$ iterations for most single-revolution cases. Robustness can be compromised near branch points but is restored with improved bracketing (Sangrà et al., 2021).

Householder Third-Order Method (Izzo)

• Employs an empirical initial guess covering the full domain and transitions to Battin/Lagrange expansions near $\chi\to 0$.
• Third-order Householder update incorporates first, second, and third derivatives:

$$\chi_{k+1} = \chi_k - T \frac{(dt)^2 - \tfrac{1}{2}T(d^2 t)}{dt[(dt)^2 - T d^2 t] + \tfrac{1}{6}d^3 t d^2 t},$$

with $T = \Delta t(\chi_k) - \Delta t_{prescribed}$.

• Converges in $2$–$3$ iterations for machine-precision accuracy, robust essentially everywhere in the single-revolution region (Sangrà et al., 2021).

Comparative Summary

Method	Typical Iterations	Robustness	Accuracy
Bate NR	5–10	Sensitive at branch pts	$<10^{-12}$ s
Izzo House.	2–3	Uniform, unconditional	$<10^{-12}$ s

Izzo’s solver is frequently deployed when automatic execution and robustness are paramount; Bate's is preferred for minimal code size and rapid per-iteration steps.

4. Geometric and Direct Path-Equation Solution

Direct manipulation based on the conic section representation allows parameter iteration over geometric invariants. The method operates in the transfer plane and leverages the following:

• The conic path equation:

$\frac{p}{r} = 1 + e\cos\theta,$

with $p=h^2/\mu$, $e$ the eccentricity.

• Time of flight via

$$t(\theta) = \sqrt{\frac{p^3}{\mu}} \int_{0}^{\theta} \frac{d\theta}{(1+e\cos\theta)^2}.$$

• For $e<1$, transformation to eccentric anomaly variable $y$ yields

$$t(\theta) = \sqrt{\frac{p^3}{\mu}} \frac{y - e\sin y}{(1-e^2)^{3/2}}.$$

• Numerical root-finding (secant or bisection) on $p$ or $a$ establishes the orbit satisfying the time-of-flight constraint, with closed-form formulas for all orbit elements and velocity vectors (Gatland, 2021, Easton et al., 2021).

5. Velocity Recovery via Lagrange’s f–g Formalism

Once orbital parameters are determined, initial and final velocities are computed using Lagrange’s coefficients:

• $f = 1 - (r_2/a)(1-\cos\Delta\theta)$,
• $g = (r_1 r_2 \sin \Delta\theta)/\sqrt{\mu a}$,
• $\dot{g} = 1 - (r_1/a)(1-\cos\Delta\theta)$.

The velocities are

$$\mathbf{v}_1 = \frac{\mathbf{r}_2 - f\mathbf{r}_1}{g},\qquad \mathbf{v}_2 = \frac{\dot{g}\mathbf{r}_2 - \mathbf{r}_1}{g}.$$

These satisfy the two-point boundary conditions and the orbital energy integral. Polar components can also be obtained directly in the orbital frame, rotated back to inertial space as required (Easton et al., 2021).

6. Treatment of Special and Degenerate Cases

Algorithms handle several pathological scenarios:

• $\theta\to 0$ (radial): $A\to 0$, yields special series.
• Parabolic limit ($\chi\to 0$): explicit expansions in Stumpff functions or switching to Battin/Lagrange z-series.
• Antipodal transfer ($\theta=\pi$): Gauss f/g method singular, use r–t formula.
• Multi-revolution: equations admit multiple valid roots for $\chi$ or $a$, each branch corresponding to distinct solution topologies (Sangrà et al., 2021).

7. Applications and Implementation Considerations

Lambert guidance is integral to orbit design, trajectory optimization, real-time guidance, and maneuver planning in modern GNC architectures. The universality of the solution applies to orbit rendezvous, interception, space debris correlation, and targeting applications. Algorithms are implemented in platforms such as ESA’s PyKEP and other embedded guidance systems where both speed and absolute reliability in convergence are requirements (Sangrà et al., 2021). Worked examples with explicit parameter values are documented for benchmarking and validation (Gatland, 2021).

• "Review of Lambert's problem" (Sangrà et al., 2021)
• "Gravitational Orbits and the Lambert Problem" (Gatland, 2021)
• "An elementary solution to Lambert's problem" (Easton et al., 2021)